Residual Plot against Fits - Is this ok?

Nash

New Member
#1
Hi!

So I've done an experiment and had some data to analyse and I was looking at all the various residual plots situation to see where my data might be far from ideal and I know that I potentially have two outliers as can be seen from top left and bottom left graphs.

However, Im not sure about top right graph which is residuals vs fits Because it is all technically random around the 0, but I can't not see the clear pattern in the plot. Is that still all random noise or is there some bias and stuff going on?

I would appreciate any suggestions, even if intiutive. Thanks!
 

ted00

New Member
#3
OP is right about that top right graph, definitely appears to be bias, but as GretaGarbo said, it's fine
I'll just add, it's fine if you're doing population-level inference
I've only seen patterns like this with longitudinal data where inference was focused on individuals
not sure which is the case in your study
Also, there's exactly 2 outliers at +/-5 having opposite signs? What kind of data is this? Can you show us a plot of the dependent variable versus the primary independent variable? Is this data on 2 subjects?
 

staassis

Active Member
#4
Slight non-normality at the tails and potential heteroskedasticity. Formal tests are needed to verify the conjecture.